The bioengineering approach utilizes preformed scaffolds, which are either chemically engineered, such as polylactic acid (PLA), polyglycolic acid (PGA), and polyglycerol sebacate (PGS) [4, 5, 9], or derived from biological sources, such as collagen/gelatin from tendon, alginates from seaweed, and silk from silkworms or spiders [27, 30, 42], and subsequently processed to generate sponges or meshes with various pore sizes. Cells are eventually seeded onto these scaffolds and cultured in a three-dimensional (3D) format. Importantly, most of the above-mentioned materials, with collagen and silk being exceptions, do not support cardiomyocyte attachment or growth in an optimal way. In addition, limited diffusion and, thus, oxygen and nutrient supply have been main caveats of the bioengineering approach. Moreover, synthetic polymers commonly have unfavorable degradation properties (e.g., the acidification of surrounding tissue in the case of PLA and PGA) and cause inflammatory responses [52]. This is acceptable or even desirable if the above-mentioned materials are used as suture material, but it is unacceptable if the materials are used as tissue scaffolds in vivo. To overcome this limitation, one might opt to degrade the respective materials before implantation. From the engineering and regulatory point of view, the above-mentioned (bio-)materials would be advantageous because its components are fully chemically defined and have already been approved (e.g., PLA, PGA, alginate, collagen/gelatin sponges) by the US Food and Drug Administration (FDA).

The biological assembly approach takes advantage of the intrinsic propensity of cardiomyocytes to generate “cardiac microtissues” if cultured at a high cell density in suspension [1, 21, 22, 34]. Suspending cells in casting molds containing mixtures of extracellular matrix material, such as collagen, laminin, and fibronectin [10, 36, 56, 57], or naturally occurring products from blood, such as fibrin [19], supports self-aggregation in a defined 3D environment. Exposing the developing tissues to biophysical stimuli, such as mechanical strain and electrical stimulation [12, 42], can further be used to guide cardiac tissue formation and will eventually yield strongly contracting heart tissue equivalents [57, 59]. In contrast to the bioengineering approach, the shape and size of forming tissues are not governed by the matrix material itself, but by the casting molds utilized for initial tissue reconstitution. Although diffusion is also a limitation of the biological assembly approach, it appears that hydrogels enable better oxygen and nutrient transport as compared to preformed matrixes. This might be the reason why tissue formation in, for example, engineered heart tissue (EHT) is not restricted to outer layers of the collagen hydrogel-culture format [57]. Instead, muscle within EHTs forms a delicate 3D network, composed of 20–200-μm-thick muscle bundles (Fig. 2). Importantly, single-muscle units generated by the biological assembly approach might be fused to form large tissue constructs [36]. This property might in principle even be exploited to generate clinical-scale tissue constructs of various geometries.

The cell sheet approach is based on temperature-controlled release of cell monolayers from PIPAAm (polyisopropylacrylamide)-coated culture dishes [38]. PIPAAm has unique hydrophobic or hydrophilic properties depending on the environmental temperature. Physiological temperatures (37°C) facilitate cell attachment, whereas low temperatures (20°C) render the polymeric surface hydrophilic and consequently lead to rapid detachment of cell monolayers. Importantly, this detachment procedure does not disturb cell–cell contacts within the monolayer and apparently maintains all cell surface structures. This might explain why cell sheets can rapidly attach and establish electrical contacts to each other [18, 47] as well as to underlying myocardium if applied in vivo [31]. Yet, the cell sheet approach is also diffusion limited. This becomes apparent at a monolayer-stack diameter of ~100 μm (four to six layers) [46]. Employing sequential grafting of thin layers on top of each other can, however, be exploited to generate compact vascularized tissues with a diameter of at least 1 mm in vivo [47].

The decellularization-recellularization approach employs tenside (SDS: sodium dodecyl sulfate) and DNase treatment to dispose of all potentially immunogeneic structures of the heart and subsequent reseeding of the remaining extracellular matrix (ECM) with cells either by direct injection into the “naked” ECM or transfusion through spared vascular channels [39]. The latter approach appears better suited to reestablish a homogenous musculature in the heart but would require that cells migrate into the heart’s ECM and form a continuous functional syncytium. Future studies will have to provide evidence that this can indeed be achieved and that the resulting hearts can be transplanted and perform as desired in vivo. To this end, first short-term data from a heterotopic heart transplantation model are encouraging [39].

Human heart cells can be isolated from biopsy and autopsy material. These cells will, however, not be suitable for myocardial tissue engineering given their limited number, lack of cell cycle activity, and inability to reestablish a functional syncytium once they have been dispersed into single-cell suspensions. Consequently, cardiogenic stem cells will have to be exploited. Adult stem cells from the bone marrow do not harbor the capacity to generate the substantial amount of myocytes needed for complex myocardial reconstitution [3, 35, 37]. In contrast, embryonic stem cells can, in principle, provide cardiomyocytes in adequate numbers [55]. In addition, embryonic stem-cell-derived myocytes are still capable of developing into complex functional syncytia [20] and, only recently, the use of embryonic stem-cell-derived cardiomyocytes was demonstrated in myocardial tissue engineering [6, 17]. Collectively, these data provide strong evidence that embryonic stem cells might be an appropriate cell source for clinical-scale myocardial tissue engineering. However, cardiomyocyte yield from embryonic stem cells is generally low (1–5%; [23]) and scalability will, consequently, be an important issue to generate relevant cardiomyocyte numbers and eventually “force-generating” human myocardium. The later goal has, despite two recent reports using embryonic stem-cell-derived myocytes in a 3D culture format [6, 45], not been achieved, but it is likely that hydrogel-cultures will be instrumental to generate such tissues [60]. Similarly, nonembryonic pluripotent stem cells, including induced pluripotent stem cells [49], spermatogonial stem cells [16], and parthenogenetic stem cells [51], will be exploitable in cardiac tissue engineering. These cell types will, on the one hand, not require cell harvest from an early embryo and might, in addition, be suitable for the generation of autologous cells; however, other ethically controversial issues remain, including the need for genetic manipulations in the induced pluripotent stem-cell-based technology and the risk of in vitro mutagenesis and teratoma formation after an in vivo application.

It will be essential to generate and implant thick muscle tissues to achieve meaningful contractile support in the setting of heart failure, to adequately replace scarred heart tissue in individuals with myocardial infarction, or to repair tissue deficiencies in pediatric hearts. The need for continuous oxygen and nutrient supply to metabolically highly active myocardium is a key issue for all tissue-engineering technologies. Although myocardium can already be reconstituted in vivo with a diameter of up to 1 mm [47, 59], this would likely be insufficient to impart a clinically relevant effect. In vivo physiological muscle diameters of at least 10 mm need to be matched. To achieve this goal, “in vitro vascularization” might be a prerequisite. It is unlikely that this will be achieved by simply adding endothelial cells or smooth muscle cells to a tissue reconstitution mixture, although it has been clearly documented that nonmyocytes will have beneficial effects, which include the formation of primitive capillary structures in tissue-engineered heart muscle [28, 36, 57]. This will, however, not likely be sufficient to provide immediate circulatory support to large engineered tissue grafts in vivo. To overcome this limitation, vascularized myocardial tissues could be generated in vivo either by seeding cardiogenic cells in prevascularized chambers [33] or by introducing a macrovasculature with defined inflow and outflow to enable surgical anastomoses to, for example, the coronary arteries [24].

Tissue-engineered myocardium would ideally be generated from biocompatible and autologous material. Biocompatible matrix materials are commonly available (e.g., collagen). Unfortunately, autologous cardiogeneic cells are not readily available. Adult stem cells from the blood or bone marrow would be logistically ideal but do not have the intrinsic capacity to give rise to cardiomyocytes of sufficient quality and quantity [2]. In contrast, embryonic stem cells can give rise to bona fide cardiomyocytes [8, 20, 54], which would, however, elicit an immune response in an allogenic recipient [25]. Alternatively, induced pluripotent [49], spermatogonial [16], and parthenogenetic [51] stem cells could be applied autologously. These cell types appear to have the most characteristics of embryonic stem cells, including the capacity to give rise to cardiomyocytes. This makes them an attractive source for cardiac-muscle-engineering and other tissue-engineering applications. Finally, the necessity for an autologous application needs to be carefully considered, given the time needed to generate a therapeutic autologous cell pool (weeks to months) and, subsequently, an autologous tissue (takes roughly 7 days when embryonic stem cells are used to generate EHTs; our own unpublished data). Having this in mind, it might be worthwhile to consider cell banking for tissue-engineering applications. In this scenario, cells with defined immunological features would be banked to be matched with an allogeneic recipient. Under these circumstances, it would be likely that induction of differentiation followed by cardiac muscle engineering could be achieved within 2–3 weeks. This time frame appears clinically acceptable. In addition, palliative procedures could bridge the gap until tissue-engineered products would be ready for a clinical application. Time-frame concerns for early postnatal heart surgery might also be irrelevant if immune-matched heart tissue could be generated to be available at birth.